An automated ticketing system for aerospace science popularization venues

Abstract

This invention relates to the technical field of ticketing and inspection, and discloses an automatic ticketing and inspection system for aerospace science popularization venues. The system includes a venue time slot capacity allocation module, a ticket waiting list establishment module, a gate passenger flow prediction module, and a gate entrance / exit direction scheduling module. First, time slot capacity is allocated according to demand-capacity matching rules to obtain a set of visitor allocation data. Second, a ticketing elastic scaling mechanism is introduced to expand the visitor allocation data set in stages, and then visitor priority scores are calculated to establish a ticket waiting list. Next, a multimodal spatiotemporal LSTM passenger flow prediction model is established by fusing time series and spatial topology to output gate passenger flow prediction values. Finally, an improved beluga optimization algorithm is used to determine the gate direction, and a gate elastic strategy is introduced to dynamically adjust the gate entrance / exit direction to complete automatic ticketing. This invention achieves the purpose of automatic ticketing and inspection by processing and modeling visitor data, and the method is objective and accurate.

Description

Technical Field

[0001] This invention relates to the technical field of ticket sales and inspection, specifically to an automatic ticket sales and inspection system for aerospace science popularization venues. Background Technology

[0002] With the rapid development of intelligent technologies, automated ticketing systems have been widely applied in transportation, entertainment, and scenic areas, significantly improving ticketing efficiency and user experience. Current mainstream systems integrate biometrics (such as facial recognition and fingerprint recognition), dynamic QR codes, contactless payment (such as NFC and UnionPay QuickPass), and cloud computing technologies, gradually achieving paperless and contactless services. However, due to the unique characteristics of aerospace science museums (such as large fluctuations in instantaneous visitor numbers, dynamic ticketing rules, and high demands for immersive experiences), existing general-purpose automated ticketing solutions still have significant room for improvement in terms of adaptability, interactivity, and intelligence.

[0003] Traditional automated ticketing methods typically involve users purchasing tickets online, where the system generates static electronic tickets (such as QR codes or digital IDs) and binds them to their identity information. Upon entry, the gate verifies the validity of the ticket using a barcode scanner or card reader, while also relying on a local database or cloud interface for real-time verification. The core process follows a fixed, unidirectional linear logic of "ticket purchase - verification - release".

[0004] Traditional automated ticketing methods require visitors to make reservations for aerospace exhibits (such as space capsule simulation experiences) in time slots. These methods rely on pre-programmed fixed rules, making it difficult to respond to dynamic adjustments in real time. During peak visitor volumes, relying on unidirectional linear logic can easily lead to entrance congestion, and the instantaneous high-concurrency processing capacity is insufficient. Summary of the Invention

[0005] In view of the problems in the related technologies, the present invention provides an automatic ticketing system for aerospace science popularization venues to overcome the technical problems existing in the prior art.

[0006] To solve the aforementioned technical problem, the present invention is achieved through the following technical solution:

[0007] This invention is an automatic ticketing system for aerospace science popularization venues, specifically including: a venue time slot capacity allocation module, a ticket waiting queue establishment module, a gate passenger flow prediction module, and a gate entrance and exit direction scheduling module;

[0008] The venue time slot capacity allocation module is used to collect visitor data of aerospace science popularization venues, allocate time slot capacity according to demand-capacity matching rules, and obtain a visitor allocation data set;

[0009] The ticket waiting list establishment module is used to introduce a ticketing elastic scaling mechanism to expand the tourist allocation data set in stages, calculate the tourist priority score to start the ticket waiting list, and realize automatic ticketing.

[0010] The gate passenger flow prediction module is used to establish a multimodal spatiotemporal LSTM passenger flow prediction model after automatic ticketing by integrating time series and spatial topology, and output the gate passenger flow prediction value.

[0011] The gate entrance / exit direction scheduling module is used to establish a multi-objective function based on the predicted gate passenger flow and the maximum passage speed, use the improved Whale Optimization Algorithm to determine the gate entrance / exit direction, and dynamically adjust the optimized gate entrance / exit direction.

[0012] Preferably, the collection of visitor data for aerospace science popularization venues includes:

[0013] Collect visitor data for aerospace science popularization venues to obtain a set of visitor numbers, number each area of ​​the aerospace science popularization venues to obtain a set of visitor areas, and calculate the basic capacity of each visitor area.

[0014] Preferably, the time-period capacity allocation according to the demand-capacity matching rule includes:

[0015] When the basic capacity of a tourist area is less than the number of tourists in the corresponding tourist area in the tourist quantity set, the demand-capacity matching rule is triggered to calculate the tourist expansion capacity and tourist carrying capacity.

[0016] When the number of tourists in the tourist count set is greater than the tourist carrying capacity, it is a peak period. When the sum of the number of two adjacent tourists in the tourist count set is less than half of the tourist carrying capacity, it is a low period, forming a new tourist area period.

[0017] Based on the new tourist area time periods, the operating hours of tourist areas are divided, and combined with the tourist number set, a tourist allocation data set is obtained.

[0018] Preferably, the step-by-step expansion of the tourist allocation dataset by introducing a flexible ticketing mechanism includes:

[0019] The tourist allocation dataset is expanded in stages: when the tourist area time period in the tourist allocation dataset is a peak period, the tourist area time period is shortened; when the tourist area time period in the tourist allocation dataset is a low period, the tourist area time period is increased; a quantity warning threshold is set: when the number of tourists in the tourist area of ​​the tourist allocation dataset exceeds the quantity warning threshold, a tourist backup area is activated.

[0020] Preferably, the step of calculating tourist priority scores to initiate the ticket waiting list includes:

[0021] Based on visitor data from aerospace science popularization venues, visitor priority scores are calculated and sorted in descending order to obtain an initial ticket waiting list. A dynamic ticketing mechanism is then established to adjust the initial ticket waiting list in real time, thereby achieving automatic ticketing.

[0022] Preferably, the method of fusing time series data and spatial topology to establish a multimodal spatiotemporal LSTM pedestrian flow prediction model includes:

[0023] After automatic ticketing, data from the aerospace science popularization venue is acquired at a fixed time granularity, normalized, and then the gate coordinates are acquired to construct a data vector and form a ticket checking data set.

[0024] Data from historical aerospace science popularization venues were collected again at a fixed time granularity to obtain a set of ticket checking data samples. The set of ticket checking data samples was then divided into a training set and a test set.

[0025] The multimodal spatiotemporal LSTM is configured to include an input layer, spatiotemporal joint encoding, cross-modal feature fusion, and an output layer, with a multi-task weighted loss function. The training and test sets are input into the multimodal spatiotemporal LSTM for training. The spatiotemporal joint encoding includes a graph convolutional network and a bidirectional LSTM network, which are transmitted to the cross-modal feature fusion layer to control the contribution ratio of spatiotemporal features. The output layer outputs the pedestrian flow prediction results, resulting in a multimodal spatiotemporal LSTM pedestrian flow prediction model.

[0026] Preferably, the predicted pedestrian flow value for the output turnstile includes:

[0027] The ticket checking data set is input into the multimodal spatiotemporal LSTM pedestrian flow prediction model, and the future pedestrian flow at each gate entrance and exit is output to obtain the gate pedestrian flow prediction value.

[0028] Preferably, determining the gate entrance / exit direction using the improved beluga optimization algorithm includes:

[0029] The gate entrance and exit directions are set according to the predicted passenger flow value, and a multi-objective function is established based on the maximum passage speed.

[0030] The beluga optimization algorithm is improved by introducing a balance factor and a nonlinear adaptive parameter, resulting in an improved beluga optimization algorithm; the multi-objective function is used as the fitness function.

[0031] Assume that a beluga whale population exists in the search space, and treat individual beluga whales in the population as the entrance and exit directions of the turnstiles, and initialize the beluga whale population;

[0032] A new balancing factor is introduced to replace the traditional balancing factor. The location of the beluga whales is updated based on whether the beluga whale population has entered the exploration or development stage.

[0033] The beluga whale falling probability is replaced with a nonlinear adaptive parameter, and the beluga whale position is updated again. The beluga whale population is updated and the next iteration begins, until the current iteration count reaches the maximum iteration count, at which point the iteration stops, and the final beluga whale population is obtained. In the final beluga whale population, the beluga whale individual corresponding to the best fitness function is found to obtain the optimized gate entrance / exit direction.

[0034] Preferably, the dynamically adjusted and optimized gate entrance / exit direction includes:

[0035] The optimized gate entrance / exit direction is used as the current gate entrance / exit direction. The gate is dynamically adjusted using an elastic strategy. The gate that is closest to the current gate and is in closed mode is found. The gate entrance / exit direction is set to be consistent with the current gate entrance / exit direction. One gate in closed mode is reserved as a backup gate.

[0036] The present invention has the following beneficial effects:

[0037] 1. This invention allocates time slot capacity according to the demand-capacity matching rule, takes into account the venue's capacity, matches the number of tourists according to the optimal tourist carrying capacity based on reservation data, and divides the operating hours of tourist areas so that tourists can effectively avoid congestion in areas such as time slot reservations, and ensure stable operation when there is no sudden flow of traffic.

[0038] 2. This invention introduces a flexible ticketing mechanism for tiered capacity expansion, refines dynamic range expansion, enables rapid response to large passenger flows during peak hours, reduces losses from frequent adjustments during off-peak hours, and dynamically sorts the waiting queue by calculating tourist priority scores. This overcomes the drawbacks of relying on pre-programmed candidate rules, avoids tourists being unable to make reservations due to full capacity, and balances the rights of early-registered users with system stability.

[0039] 3. This invention establishes a multimodal spatiotemporal LSTM pedestrian flow prediction model by fusing time series and spatial topology to predict pedestrian flow at gate entrances and exits. This neural network combines multimodal data fusion, spatiotemporal dependency modeling, and dynamic decision response to overcome the limitations of traditional neural networks in real-time traffic prediction (such as ignoring spatiotemporal correlation), greatly improving prediction accuracy. The minute-level prediction response meets the needs of dynamic scheduling, providing a reliable theoretical and practical foundation for intelligent venue management of highly dynamic pedestrian flow.

[0040] 4. This invention determines the gate direction by using an improved whale optimization algorithm and introduces a gate elasticity strategy to dynamically adjust the gate entrance and exit directions, thus completing automatic detection. This method quickly solves the gate direction switching problem in a multi-objective optimization model, introduces a balance factor and nonlinear adaptive parameters to improve the whale optimization algorithm, achieves a balance between exploration and development, effectively avoids getting trapped in local optima, and improves the efficiency of the algorithm in local search. Combined with the predicted flow of the entrance and exit, the gate direction is dynamically adjusted to reduce the average waiting time for users, reduce the probability of gate entrance and exit congestion, and improve the instantaneous high-concurrency processing capability.

[0041] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0042] To more clearly illustrate the technical solutions of the embodiments of the invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, the drawings can be obtained from these drawings without creative effort.

[0043] Figure 1 The present invention provides a flowchart of an automatic ticketing system for aerospace science popularization venues.

[0044] Figure 2 This invention provides a flowchart illustrating an automatic ticketing method for aerospace science popularization venues. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] Traditional automated ticketing methods require visitors to make reservations for aerospace exhibits (such as space capsule simulation experiences) in time slots. These methods rely on pre-programmed fixed rules, making it difficult to respond to dynamic adjustments in real time. During peak visitor volumes, relying on unidirectional linear logic can easily lead to congestion at entrances and exits, and the system lacks the capacity to handle high concurrency in an instant.

[0047] To solve the above technical problems, such as Figure 1As shown in the figure, this embodiment of the invention provides an automatic ticketing system for aerospace science popularization venues, specifically including: a venue time slot capacity allocation module, a ticket waiting queue establishment module, a gate passenger flow prediction module, and a gate entrance / exit direction scheduling module; the venue time slot capacity allocation module is used to collect visitor data of aerospace science popularization venues, allocate time slot capacity according to demand-capacity matching rules, and obtain a visitor allocation data set; the ticket waiting queue establishment module is used to introduce a ticketing elastic scaling mechanism to expand the visitor allocation data set in stages, calculate visitor priority scores to start the ticket waiting queue, and realize automatic ticketing; the gate passenger flow prediction module is used to establish a multimodal spatiotemporal LSTM passenger flow prediction model by fusing time series and spatial topology after automatic ticketing, and output the gate passenger flow prediction value; the gate entrance / exit direction scheduling module is used to establish a multi-objective function based on the gate passenger flow prediction value and the maximum passage speed, use an improved white whale optimization algorithm to determine the gate entrance / exit direction, and dynamically adjust the optimized gate entrance / exit direction.

[0048] A specific implementation involves deploying intelligent turnstiles (such as double-door structures with dynamic indicator lights) that support real-time switching of passage direction in aerospace science popularization venues. The entrance / exit functions are dynamically allocated through a central control system. Real-time crowd density data is collected through built-in cameras, infrared sensors, and venue Wi-Fi probes in the turnstiles, enabling real-time crowd perception and providing high-quality data support for this invention.

[0049] In the specific implementation process of the above embodiments, firstly, visitor data of aerospace science popularization venues is collected, and time-slot capacity is allocated according to demand-capacity matching rules to obtain a visitor allocation data set. This method considers the venue's capacity, matches the number of visitors according to the optimal visitor carrying capacity based on reservation data, and divides the operating hours of visitor areas to effectively avoid congestion in areas such as time-slot reservations, ensuring stable operation when there is no sudden flow of traffic. Secondly, a ticketing elastic scaling mechanism is introduced to expand the visitor allocation data set in stages, and then the visitor priority score is calculated to start the ticket waiting queue and realize automatic ticketing. This method refines the dynamic interval expansion, quickly responds to the instantaneous large flow of people during peak hours, and reduces the loss of frequent adjustments during off-peak hours. At the same time, the waiting queue is dynamically sorted by visitor priority score to overcome the drawbacks of relying on pre-programmed candidate rules, avoids visitors being unable to make reservations due to full capacity, and takes into account the rights of early registration users and system stability. Then, a multimodal spatiotemporal LSTM traffic prediction model is established by acquiring aerospace science popularization venue data and integrating time series and spatial topology to predict the traffic flow at the gate entrance and exit. The method performs predictions and outputs the gate passenger flow forecast. This neural network combines multimodal data fusion, spatiotemporal dependency modeling, and dynamic decision response to overcome the limitations of traditional neural networks in real-time traffic prediction (such as ignoring spatiotemporal correlation), greatly improving prediction accuracy. The minimum prediction response time is in the minute range, meeting dynamic scheduling requirements and providing a reliable theoretical and practical foundation for intelligent venue management with high dynamic passenger flow. Finally, based on the gate passenger flow forecast and a multi-objective function established based on the maximum passage speed, the improved Whale Optimization Algorithm is used to determine the gate direction. A gate elasticity strategy is introduced to dynamically adjust the gate entrance and exit directions, completing automatic ticket checking. This method quickly solves the gate direction switching problem in a multi-objective optimization model, introduces a balance factor and nonlinear adaptive parameters to improve the Whale Optimization Algorithm, achieving a balance between exploration and exploitation, effectively avoiding getting trapped in local optima, and improving the algorithm's efficiency in local searches. Combining the predicted flow at entrances and exits, the gate direction is dynamically adjusted, reducing the average user waiting time, decreasing the probability of gate entrance and exit congestion, and improving instantaneous high-concurrency processing capabilities.

[0050] Furthermore, to better illustrate the technical solution of the embodiments of the present invention, based on the above-mentioned automatic ticketing system for aerospace science popularization venues, such as... Figure 2 As shown in the figure, this embodiment of the invention provides an automatic ticketing method for aerospace science popularization venues, specifically including the following:

[0051] S1. Collect visitor data for aerospace science popularization venues to obtain a visitor number set, and allocate time-period capacity to the visitor number set according to the demand-capacity matching rule to obtain a visitor allocation data set;

[0052] S1 includes the following steps:

[0053] S11. Collect visitor data for the aerospace science popularization venue. The visitor data includes the capacity of each area of ​​the aerospace science popularization venue, visitor registration time, visitor stay time, number of no-shows, etc. Calculate the number of visitors in each area of ​​the aerospace science popularization venue to obtain a set of visitor numbers. Set demand-capacity matching rules, number each area of ​​the aerospace science popularization venue to obtain a set of visitor areas, obtain the maximum capacity of each visitor area in the set of visitor areas, set a visitor capacity coefficient, calculate the product of the maximum capacity of the visitor area and the visitor capacity coefficient, and calculate the basic capacity of the visitor area.

[0054] S12. Allocate time-period capacity to the tourist quantity set according to the demand-capacity matching rule to obtain the tourist allocation data set. The specific steps are as follows:

[0055] S121. When the basic capacity of a visitor area is less than the number of visitors in the corresponding visitor area in the visitor count set, the demand-capacity matching rule is triggered, and the visitor capacity expansion is calculated. Where C max This indicates the maximum capacity of the visitor area. C represents the number of visitors in the visitor area, and C represents the basic capacity of the visitor area.

[0056] S122. Implement safety limits on the basic capacity and expansion capacity of tourist areas, and introduce a safety factor to ensure tourist carrying capacity. Where γ represents the safety factor, which is a constant in practice and can be adjusted periodically according to management objectives and actual operating data; the operating hours of the tourist area are divided according to fixed time intervals to obtain the initial time period of the tourist area, and then the time period capacity is allocated;

[0057] For any time period in a tourist area, when the number of tourists in the tourist quantity set is greater than the tourist carrying capacity, it is a peak period; when the sum of the number of two adjacent tourists in the tourist quantity set is less than half of the tourist carrying capacity, it is a low period. At this time, the peak period and the low period are marked to form a new tourist area time period; otherwise, the initial time period of the tourist area is maintained.

[0058] S123. Divide the operating hours of the tourist areas again according to the new tourist area time periods, and combine them with the tourist number set to obtain the tourist allocation data set;

[0059] In this embodiment, visitor data for aerospace science popularization venues is collected, and time-slot capacity is allocated according to demand-capacity matching rules to obtain a visitor allocation data set. This method considers the venue's capacity, matches the number of visitors based on reservation data according to the optimal visitor carrying capacity, and divides the operating hours of visitor areas to effectively avoid congestion in areas such as time-slot reservations, ensuring stable operation when there are no sudden surges in traffic. Specifically, for example, area A has a maximum capacity of 200 visitors, area B has a maximum capacity of 150 visitors, and visitor registration data are 180 and 100 respectively. Area A has a basic capacity of 160 visitors, and area B has a basic capacity of 120 visitors. In area A (180 > 160), capacity expansion is triggered, and the calculated expansion capacity is 20. In area B (100 < 120), no expansion is needed. For example, a safety factor of 0.9 is used, reserving a 10% flexibility. Redundancy is used to cover most of the error range, ensuring that the lower limit of the actual allocated capacity can still cover 90% of the expected demand, avoiding resource shortages due to overestimation in forecasts; the safe carrying capacity of area A is 162, and the safe carrying capacity of area B is 108; divided into time slots of 2 hours each, time slot 1 (9:00-11:00), time slot 2 (11:00-13:00), time slot 3 (13:00-15:00), and time slot 4 (15:00-17:00); the number of visitors in time slot 1 of area A is 170 > 162 → peak period; the number of visitors in time slot 2 of area A is 40; the number of visitors in time slot 3 is 30; the sum of the two adjacent time slots is 70 < 81 → low period periods are merged; the number of visitors in time slots 1-4 of area B is all < 108 → the original time slots are maintained; the entire process optimizes visitor distribution through time slot marking, balancing safety and experience;

[0060] S2. Introduce a ticketing elastic scaling mechanism to expand the tourist allocation data set in stages, then calculate the tourist priority score, and start the ticketing waiting queue according to the tourist priority score to realize automatic ticketing;

[0061] S2 includes the following steps:

[0062] S21. The tourist allocation data set includes tourist area time period and tourist number in tourist area. A flexible ticketing expansion mechanism is established in combination with spatiotemporal dimensions to expand the tourist allocation data set in stages.

[0063] When the tourist area time period in the tourist allocation dataset is a peak period, the tourist area time period is further divided, shortening the tourist area time period in the time dimension; when the tourist area time period in the tourist allocation dataset is a low period, the tourist area time periods are merged, increasing the tourist area time period in the time dimension; a quantity warning threshold is set, and when the number of tourists in the tourist area of ​​the tourist allocation dataset exceeds the quantity warning threshold, a tourist backup area is activated to complete the spatial dimension gradient expansion;

[0064] S22. After the tiered expansion is completed, calculate the tourist priority score and start the ticket waiting queue according to the tourist priority score. The specific steps are as follows:

[0065] S221. Obtain the visitor registration time and the number of no-shows from the visitor data of the aerospace science popularization venue, and obtain the number of days the visitor has been waiting since the visitor's registration time; set the priority weights as the first priority weight ω1 and the second priority weight ω2 respectively, and at this time the visitor priority score B = ω1·t1-ω2·t2, where t1 represents the number of days the visitor has been waiting since the visitor's registration time, and t2 represents the number of no-shows.

[0066] S222. Arrange the tourist priority scores in descending order to establish an initial ticket waiting queue, and establish a dynamic ticketing mechanism to adjust the initial ticket waiting queue in real time.

[0067] Two hours before ticket sales begin, the first priority weight is increased, and the initial ticket waiting list is updated. A no-show threshold is set. When the number of no-shows exceeds the threshold, the second priority weight is increased, and the initial ticket waiting list is updated again. When ticket sales begin, a specific time period is set. If a tourist in the initial ticket waiting list fails to confirm within the specified time period, the corresponding tourist's priority score is placed at the end of the initial ticket waiting list to obtain the final ticket waiting list, thus achieving automatic ticket sales.

[0068] In this embodiment, a flexible ticketing mechanism is introduced to expand the tourist allocation data set in stages, then tourist priority scores are calculated, and a ticket waiting list is activated to achieve automatic ticketing. This method refines the dynamic range expansion, enabling rapid response to sudden large passenger flows during peak hours and reducing losses from frequent adjustments during off-peak hours. Simultaneously, the waiting list is dynamically sorted based on tourist priority scores. Specifically, for example, during peak hours (10:00-12:00), ticketing units are divided into 30-minute segments: 10:00-10:30, 10:30-11:00, 11:00-11:30, and 11:30-12:00; off-peak hours (14:00-16:00) are merged into a single long segment: 14:00-16:00, allowing tourists flexible entry. (Area A, main exhibition hall) The warning threshold is 200 people. When the number of real-time reservations reaches 220, the backup exhibition hall in Zone B is automatically activated to divert 50 people, achieving a gradual expansion of space capacity. Visitor priority is quantified through linear combination, balancing fairness and credit management while improving visitor satisfaction. Positive incentives are given for waiting time (ω1·t1), with longer waiting times resulting in higher scores, reducing the risk of visitors leaving due to excessive waiting. Negative incentives are given for no-show behavior (-ω2·t2), with more no-shows resulting in lower scores, reducing the priority of visitors with poor credit and preventing repeated waste of resources. For example, if the number of days a visitor waits is used as a core indicator, with an initial parameter of ω1 = 0.6 to ensure the priority weight sum is 1, then no-shows are penalized with an initial parameter of ω2 = 0.4. More than one no-show is considered a sign of poor credit, with a no-show threshold of one instance.

[0069] Tourist 1 registered 10 days ago with 0 no-shows; Tourist 2 registered 5 days ago with 2 no-shows; Tourist 3 registered 8 days ago with 1 no-show. Priority scores are calculated: Tourist 1: 6.0; Tourist 2: 2.2; Tourist 3: 4.4. The initial queue order is 1-3-2. Two hours before ticket sales, the weights change (ω1 = 0.8), updating the queue to 1-3-2. No-show penalties are triggered (No-show 2 exceeds the threshold, ω2 = 0.6), and the queue remains 1-3-2. Tourist 1 completes payment within 5 minutes, securing their seat. Tourist 3, having exceeded the time limit, is moved to the back of the queue, resulting in a final queue order of 1-2-3. This process overcomes the drawbacks of relying on pre-programmed candidate rules, prevents tourists from being unable to book due to full capacity, and balances the rights of early registrants with system stability.

[0070] S3. After automatic ticketing, acquire data from the aerospace science popularization venue to obtain a set of ticket checking data. Then, integrate time series and spatial topology to establish a multimodal spatiotemporal LSTM people flow prediction model to predict the people flow at the gate entrance and exit, and output the predicted gate people flow value.

[0071] S3 includes the following steps:

[0072] S31. After automatic ticketing, acquire aerospace science popularization venue data according to a fixed time granularity. The aerospace science popularization venue data includes the flow of people at the entrance gate, the flow of people at the exit gate, the distance between the entrance gate and the exit gate, and the flow density of people in the visitor area. The flow of people at the entrance gate and the flow of people at the exit gate are used as time series data, and the distance between the entrance gate and the exit gate and the flow density of people in the visitor area are used as spatial topology data. Set the maximum carrying capacity within the fixed time granularity, normalize the flow of people at the entrance gate and the flow of people at the exit gate to the range of [0, 1], and obtain the normalized value of the flow of people at the entrance gate and the normalized value of the flow of people at the exit gate respectively. Then, establish the plane coordinates of the aerospace science popularization venue, obtain the gate coordinates, construct a data vector [normalized value of the flow of people at the entrance gate, normalized value of the flow of people at the exit gate, gate coordinates], and form a ticket checking data set.

[0073] S32. A multimodal spatiotemporal LSTM pedestrian flow prediction model is established by fusing time series and spatial topology to predict pedestrian flow at turnstile entrances and exits, and the predicted pedestrian flow value is output. The specific steps are as follows:

[0074] S321. Collect historical aerospace science popularization venue data again according to a fixed time granularity to obtain a ticket checking data sample set. Divide the ticket checking data sample set into a sample training set and a sample test set. Set the multimodal spatiotemporal LSTM to include an input layer, spatiotemporal joint encoding, cross-modal feature fusion, and an output layer. The loss function is a multi-task weighted loss. Input the sample training set into the multimodal spatiotemporal LSTM for training. The spatiotemporal joint encoding includes a graph convolutional network and a bidirectional LSTM network. After capturing the spatial and temporal dependencies of the sample training set, it is transmitted to the cross-modal feature fusion to control the contribution ratio of spatiotemporal features. The output layer outputs the traffic flow prediction result. Set a maximum number of training rounds. Stop training when the maximum number of training rounds is reached to obtain the trained multimodal spatiotemporal LSTM.

[0075] S322. Input the sample test set into the trained multimodal spatiotemporal LSTM. When the loss function does not decrease for K consecutive rounds, stop training to obtain the multimodal spatiotemporal LSTM pedestrian flow prediction model. Otherwise, adjust the weights and continue training until the loss function does not decrease for K consecutive rounds. Input the ticket checking data set into the multimodal spatiotemporal LSTM pedestrian flow prediction model and output the future pedestrian flow at each gate entrance to obtain the gate pedestrian flow prediction value.

[0076] In this embodiment, data from aerospace science popularization venues is acquired, and a multimodal spatiotemporal LSTM pedestrian flow prediction model is established by fusing time series and spatial topology to predict pedestrian flow at gate entrances and exits, outputting the predicted gate pedestrian flow value. This neural network combines multimodal data fusion, spatiotemporal dependency modeling, and dynamic decision response, overcoming the limitations of traditional neural networks in real-time traffic prediction (such as ignoring spatiotemporal correlation), greatly improving prediction accuracy, with a minimum prediction response time in minutes, meeting dynamic scheduling requirements; specifically, for example, data from aerospace science popularization venues is collected every 15 minutes. The Tiankepu venue data includes entrance / exit turnstile traffic flow, turnstile spacing (entrance A and exit B are 50 meters apart), and real-time traffic density in each area (e.g., exhibition hall 1 has a density of 0.8 people / ㎡; the maximum capacity of a single turnstile is set to 100 people / 15 minutes, the entrance flow (80 people) is normalized to 0.8, the exit flow (60 people) is normalized to 0.6, and the venue floor plan is mapped to a two-dimensional coordinate system (entrance turnstile coordinates are (0, 0), exit turnstile coordinates are (20, 15), and exhibition hall center coordinates are (10, 8)), forming a data vector.

[0077] [0.8, 0.6, (0, 0)]; The time series length is set to 24 hours (96 time steps); Graph convolutional network: with turnstiles as nodes, the node feature is historical passenger flow, and the edge weight is the reciprocal of the distance between turnstiles (the distance between entrance A and exit B is 50 meters, and the weight is 1 / 50); Bidirectional LSTM: the time window is set to 6 hours (24 time steps), the hidden layer dimension is 64, and it captures the peak and valley patterns of passenger flow (the peak at the entrance at 10 am); Cross-modal fusion: the attention mechanism is used to allocate weights, and the proportion of spatiotemporal features is dynamically adjusted (the weight of time features during peak periods is 0.7, and the weight of spatial features is 0.3); The maximum number of training rounds is 200, the early stop threshold is 10 (if the loss does not decrease after 10 consecutive rounds of testing, the system will terminate); the input is the real-time ticket checking data set, and the output is the predicted value of each turnstile for the next 4 hours (16 time steps). At this time, the prediction accuracy is 90%, and the prediction accuracy of a single LSTM neural network is 82%, which provides a reliable theoretical and practical basis for intelligent venue management of highly dynamic passenger flow;

[0078] Preferably, step S4 includes the following steps:

[0079] S41. Based on the predicted passenger flow at the turnstiles, set the turnstile entrance / exit directions, which include a closed mode, an exit mode, and an entrance mode, represented by 0, 1, and 2 respectively. Set a minimum switching interval to ensure that the switching interval between the exit and entrance turnstiles is greater than the minimum switching interval. Obtain the turnstile processing rate, divide the predicted passenger flow by the turnstile processing rate to obtain the visitor queuing time, and establish a multi-objective function based on the maximum passage speed. Where α1 and α2 are weighting coefficients. This indicates the queuing time for tourists, and u indicates the number of turnstile switching times.

[0080] S42. The multi-objective function is used as the fitness function, and the process of solving the optimal fitness function value is regarded as the process of solving the optimal multi-objective function value. The balance factor and nonlinear adaptive parameter are introduced to improve the white whale optimization algorithm, resulting in an improved white whale optimization algorithm. The improved white whale optimization algorithm is used to solve for the optimal multi-objective function value to obtain the optimized gate entrance and exit direction. The specific steps are as follows:

[0081] S421. Set up a search space. There is a beluga whale population in the search space. The number of beluga whales in the population is p. The dimension of the beluga whale population is q. Consider the individual beluga whales in the population as the gate entrance and exit directions. Consider the process of updating the position of the individual beluga whales as the process of finding the gate entrance and exit directions.

[0082] The beluga whale population is initialized with the initial gate direction as the initial position. The current fitness function value is calculated and used to evaluate the quality of the current beluga whale position. A new balance factor is introduced to replace the traditional balance factor. Let d represent a random number between (0, 1). The current iteration count is t, and the maximum iteration count is T. The new balance factor is then... Determine the beluga whale population stage based on the new balance factor;

[0083] When β > 0.5, the beluga whale population enters the exploration phase. The r-th beluga whale position is randomly selected within the population and denoted as W. r (t), where the position of the i-th beluga whale at the t-th iteration is denoted as W. i (t), d1 and d2 represent random numbers between (0, 1); when the dimension of the i-th beluga whale position is even, the position W of the i-th beluga whale in the (t+1)-th iteration is... i (t+1)=W i (t)+(W r (t)-W i (t))(1+r1)sin(2πr2), when the dimension of the i-th beluga whale's position is odd, we get W. i (t+1)=W i (t)+(W r (t)-W i (t))(1+r1)cos(2πr2);

[0084] When β ≤ 0.5, the beluga whale population enters the development phase, introducing Levy flight for predation. The jump intensity of Levy flight is χ, and the Levy flight function is Levy(q). Let d3 and d4 represent random numbers in the interval (0, 1). The optimal beluga whale position at the t-th iteration is W′(t). Simplifying the development phase of the beluga whale population, we obtain W. i (t+1)=d3·W′(t)+χ·Levy(q)·(W r (t)-Wi (t));

[0085] At this point, the current fitness function value of the individual beluga whale in the beluga whale population is calculated to obtain the current optimal beluga whale position and the current gate entrance / exit direction;

[0086] S422. As the beluga whale population enters the whale fall phase, a nonlinear adaptive parameter is used to replace the beluga whale fall probability, resulting in the nonlinear adaptive parameter. Let d5, d6, and d7 represent random numbers within the interval (0, 1), with an upper bound of l1 and a lower bound of l2 in the search space. The population step factor ε = 2p·δ. The beluga whale's descent step size is then defined. W was obtained based on the beluga whale's descent stride. i (t+1)=d5·W i (t)-d6·W r (t)+d7·W″;

[0087] At this point, all stages of the t-th iteration are completed, the beluga whale population is updated, and the next iteration begins. This continues until the current iteration count reaches the maximum iteration count, at which point the iteration stops, and the final beluga whale population is obtained. Within the final beluga whale population, the beluga whale individual corresponding to the optimal fitness function is searched to obtain the optimized gate entrance / exit direction.

[0088] S43. The optimized gate entrance / exit direction is taken as the current gate entrance / exit direction. The gate is then dynamically adjusted using a flexible strategy. The number of people queuing at the gate is monitored in real time. A congestion threshold is set. When the number of people queuing at the current gate exceeds the congestion threshold, the gate that is closest to the current gate and is in closed mode is found. The gate entrance / exit direction is set to be consistent with the current gate entrance / exit direction. One gate in closed mode is reserved as a backup gate to complete automatic ticket checking.

[0089] In this embodiment, based on the predicted passenger flow at the turnstiles and a multi-objective function established based on the maximum passage speed, an improved Whale Optimization Algorithm is used to determine the turnstile direction. A flexible turnstile strategy is introduced to dynamically adjust the entrance and exit directions, completing automatic ticket checking. This method quickly solves the turnstile direction switching problem in a multi-objective optimization model. A balance factor and nonlinear adaptive parameters are introduced to improve the Whale Optimization Algorithm, achieving a balance between exploration and development, effectively avoiding getting trapped in local optima, and improving the algorithm's efficiency in local searches. Combined with the predicted flow at the entrance and exit, the turnstile direction is dynamically adjusted, reducing the average user waiting time, decreasing the probability of congestion at the turnstile entrances and exits, and improving instantaneous high-concurrency processing capabilities. Specifically, for example, eight bidirectional turnstiles are set up, initially with 4 entrances / 4 exits, and each turnstile has a processing rate of 30 people / minute. The system predicts that the passenger flow distribution at each turnstile within the next 30 minutes will be 120-180 people / minute at the entrance and 80-120 people / minute at the exit. The algorithm is designed with a minimum switching interval of 5 minutes and a congestion threshold of more than 20 people in the queue. Ten beluga whale individuals are set, each with 8 dimensions (corresponding to 8 turnstiles). When the balance factor β > 0.5, the beluga whale individual uses cosine updates in the 3rd dimension (odd dimension); when β ≤ 0.5, the individual uses Levy flight updates. At the 50th iteration, the adaptive parameter is 0.606, generating a step size. After 100 iterations, the algorithm outputs the optimal direction combination: 5 entrances / 3 exits. At this point, the average queuing time decreases from 8 minutes to 3.2 minutes, and the switching frequency is controlled to 2 times (initially adjusted to 1 time, with subsequent elastic adjustments of 1 time). When the queue at a certain entrance turnstile reaches 25 people, exceeding the threshold of 20, the system detects that turnstile 3 (1.5 meters away) is in closed mode and activates turnstile 3 as the entrance direction, keeping turnstile 7 as a backup. The entire process achieves intelligent turnstile scheduling and automatic ticket checking through a three-stage closed-loop control of prediction, optimization, and adjustment.

[0090] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0091] The preferred embodiments of the invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. An automatic ticketing system for aerospace science popularization venues, characterized in that, include: The venue time slot capacity allocation module is used to collect visitor data for aerospace science popularization venues, allocate time slot capacity according to demand-capacity matching rules, and obtain a visitor allocation data set. The ticket waiting list creation module is used to introduce a ticketing elastic scaling mechanism to expand the tourist allocation data set in stages, calculate the tourist priority score to start the ticket waiting list, and realize automatic ticketing. The turnstile passenger flow prediction module is used to establish a multimodal spatiotemporal LSTM passenger flow prediction model after automatic ticketing by integrating time series and spatial topology, and output the predicted value of the turnstile passenger flow. The gate entrance / exit direction scheduling module is used to determine the gate entrance / exit direction based on the predicted gate passenger flow value and the maximum passage speed by establishing a multi-objective function, using the improved Whale Optimization Algorithm, and dynamically adjusting the optimized gate entrance / exit direction. The time-period capacity allocation according to the demand-capacity matching rule includes: When the basic capacity of a tourist area is less than the number of tourists in the corresponding tourist area in the tourist quantity set, the demand-capacity matching rule is triggered to calculate the tourist expansion capacity and tourist carrying capacity. When the number of tourists in the tourist count set is greater than the tourist carrying capacity, it is a peak period. When the sum of the number of two adjacent tourists in the tourist count set is less than half of the tourist carrying capacity, it is a low period, forming a new tourist area period. Based on the new tourist area time periods, the operating hours of tourist areas are divided, and combined with the tourist number set, a tourist allocation data set is obtained; The introduction of a flexible ticketing mechanism to expand the tourist allocation dataset in stages includes: The tourist allocation dataset is expanded in stages: when the tourist area time period in the tourist allocation dataset is a peak period, the tourist area time period is shortened; when the tourist area time period in the tourist allocation dataset is a low period, the tourist area time period is increased; a quantity warning threshold is set: when the number of tourists in the tourist area of ​​the tourist allocation dataset exceeds the quantity warning threshold, a tourist backup area is activated. The method of using the improved beluga optimization algorithm to determine the gate entrance / exit direction includes: The gate entrance and exit directions are set according to the predicted passenger flow value, and a multi-objective function is established based on the maximum passage speed. The beluga optimization algorithm is improved by introducing a balance factor and a nonlinear adaptive parameter, resulting in an improved beluga optimization algorithm; the multi-objective function is used as the fitness function. Assume that a beluga whale population exists in the search space, and treat individual beluga whales in the population as the entrance and exit directions of the turnstiles, and initialize the beluga whale population; A new balancing factor is introduced to replace the traditional balancing factor. The location of the beluga whales is updated based on whether the beluga whale population has entered the exploration or development stage. The beluga whale falling probability is replaced with a nonlinear adaptive parameter, and the beluga whale position is updated again. The beluga whale population is updated and the next iteration begins, until the current iteration count reaches the maximum iteration count, at which point the iteration stops, and the final beluga whale population is obtained. In the final beluga whale population, the beluga whale individual corresponding to the best fitness function is found to obtain the optimized gate entrance / exit direction.

2. The automatic ticketing system for aerospace science popularization venues according to claim 1, characterized in that, The collected visitor data for aerospace science popularization venues includes: Collect visitor data for aerospace science popularization venues to obtain a set of visitor numbers, number each area of ​​the aerospace science popularization venues to obtain a set of visitor areas, and then calculate the basic capacity of each visitor area.

3. The automatic ticketing system for aerospace science popularization venues according to claim 1, characterized in that, The process of calculating tourist priority scores to initiate the ticket waiting list includes: Based on visitor data from aerospace science popularization venues, visitor priority scores are calculated and sorted in descending order to obtain an initial ticket waiting list. A dynamic ticketing mechanism is then established to adjust the initial ticket waiting list in real time, thereby achieving automatic ticketing.

4. An automatic ticketing system for aerospace science popularization venues according to claim 3, characterized in that, The multimodal spatiotemporal LSTM pedestrian flow prediction model, which integrates time series data and spatial topology, includes: After automatic ticketing, data from the aerospace science popularization venue is acquired at a fixed time granularity, normalized, and then the gate coordinates are acquired to construct a data vector and form a ticket checking data set. Data from historical aerospace science popularization venues were collected again at a fixed time granularity to obtain a set of ticket checking data samples. The set of ticket checking data samples was then divided into a training set and a test set. The multimodal spatiotemporal LSTM is configured to include an input layer, spatiotemporal joint encoding, cross-modal feature fusion, and an output layer, with a multi-task weighted loss function. The training and test sets are input into the multimodal spatiotemporal LSTM for training. The spatiotemporal joint encoding includes a graph convolutional network and a bidirectional LSTM network, which are transmitted to the cross-modal feature fusion layer to control the contribution ratio of spatiotemporal features. The output layer outputs the pedestrian flow prediction results, resulting in a multimodal spatiotemporal LSTM pedestrian flow prediction model.

5. An automatic ticketing system for aerospace science popularization venues according to claim 4, characterized in that, The predicted passenger flow value for the output turnstile includes: The ticket checking data set is input into the multimodal spatiotemporal LSTM pedestrian flow prediction model, and the future pedestrian flow at each gate entrance and exit is output to obtain the gate pedestrian flow prediction value.

6. An automatic ticketing system for aerospace science popularization venues according to claim 1, characterized in that, The dynamically adjusted and optimized gate entrance / exit directions include: The optimized gate entrance / exit direction is used as the current gate entrance / exit direction. The gate is dynamically adjusted using an elastic strategy. The gate that is closest to the current gate and is in closed mode is found. The gate entrance / exit direction is set to be consistent with the current gate entrance / exit direction. One gate in closed mode is reserved as a backup gate.

7. An automatic ticketing method for aerospace science popularization venues, characterized in that, The automatic ticketing system based on any one of claims 1-6 specifically includes: S1. Collect visitor data for aerospace science popularization venues to obtain a visitor number set, and allocate time-period capacity to the visitor number set according to the demand-capacity matching rule to obtain a visitor allocation data set; S2. Introduce a ticketing elastic scaling mechanism to expand the tourist allocation data set in stages, then calculate the tourist priority score, and start the ticketing waiting queue according to the tourist priority score to realize automatic ticketing; S3. After automatic ticketing, acquire data from the aerospace science popularization venue to obtain a set of ticket checking data. Then, integrate time series and spatial topology to establish a multimodal spatiotemporal LSTM people flow prediction model to predict the people flow at the gate entrance and exit, and output the predicted gate people flow value. S4. Based on the predicted passenger flow of the turnstiles and a multi-objective function established based on the maximum passage speed, the improved Whale Optimization Algorithm is used to determine the turnstile direction, thereby obtaining the optimized turnstile entrance and exit direction. A turnstile elastic strategy is introduced to dynamically adjust the optimized turnstile entrance and exit direction to complete automatic ticket checking.

Images